# System Identification (SysID) for UAVs 

This subteam focuses on System identification (SysID) with a focus on guidance,navigation, and control (GNC). SysID focuses on accurately simulating and predicting the behavior of UAVs, in particular during abnormal flight conditions such as certain maneuvers or external disturbances. SysID also sets the foundation for model-predictive control (MPC), based on accurate state estimation. SysID methods subsume classical and data-driven methods including frequency-based methods, time-domain methods range from output error estimation to neural network-based approaches. The task of SysID is to predict the system’s future states but also learning or estimating the parameters of ordinary differential equations (ODE) or even the functional terms of the ODE describing the behavior of the UAV based on input and output data (e.g. velocity, flight path angle, angle of attack, pitch rate). Ongoing research focuses on methods for parameter estimation as well as uncertainty quantification (UQ) and residual approximation  in SysID using Bayesian methods. Team members learn about the mathematical foundations of SysID based on readings of foundational literature (e.g. scientific literature), and implement methods based on simulated data of an ordinary differential equation (ODE) of a second-order dynamical system (Mass-Damper) before moving on to 3-DOF aircraft modeling and real-world 6-DOF aircraft data from our industry partner Windracers US LLC and our own collected data using a downscaled version (Sportscub). The final SysID results will be used for the digital twin simulator, used in the fixed-wing drone competition and also for SysID of the industry partner Windracers US LLC. Potential collaboration with other industry partners is also possible (Skydio etc.).
Expected activities include: 1) Literature review 2) Mathematical Formulation of the Method 3) Simulation, Data Collection and Preprocessing 4) Model Implementation for Bayesian SySiD 5) Software engineering and vehicle config publishing (e.g. log-to-parameter pipeline), 6) Simulation and sensitivity analysis. 
Researchers also have the opportunity to learn how to fly fixed-wing drones, and collect flight data through the collaboration with PURT and SATT.